What a time to be alive. Barely a year ago, advanced generative AI was the stuff of science fiction. But now, you can ask Bing to look up movie times (if it actually works) ChatGPT to write essays and cover letters, and DALL-E to make a picture of anything you ask it to. These systems have enormous potential for our society, but with them come enormous risks. But though there’s a lot to say about the ethical issues of AI or whether it will cause the downfall of civilization, I want to focus on how the anime industry is already using AI to disenfranchise creators and make art suffer overall.
To be fair, AI has been used in anime for years. Series like Demon Slayer and Fate use algorithms to generate “in-between frames”, to save time and have smoother overall animation. And of course, streaming services like Netflix and Crunchyroll have their own alrogithms to recommend you shows based on your previous watching history. The difference is that now we have generative AI, which can “draw” art from scratch that looks surprisingly human (give or take a couple extra fingers). And companies like Netflix are already using AI-generated images: their series Dog and Boy, created by Studio Wit of Attack on Titan fame, uses AI backgrounds in place of human artists.
As many have noted, this is “a slap in the face” to real human artists and animators: people who have studied and spent years of their lives perfecting their craft. But it is especially insulting if you know even a little bit about how these systems work, and the dire circumstances animators already face in this industry.
Now, I am not a computer whiz, but I have read, from people smarter than me, a little bit about how generative AI works. (To be fair, not even the top AI researchers in the world know entirely how it works!) Basically, these systems are fed massive amounts of data from across the internet, and use complex algorithms to generate text and images based on the prompts they are given. If you give an AI a prompt like, “Stephen Colbert vs. Sephiroth”, it will go through its library to find as many images as possible of these two characters, and generate a composite picture based on the features common to all these images.
I use the term “generate” instead of “create” because AI, in its present form, cannot truly create anything; it can only give an approximation from what humans have already done. Many images in AI libraries are copyrighted and were used without permission from the artists. Plagiarism is already an issue in the art community, and these corporations are making millions off other people’s work. Right now, artists are suing companies like OpenAI and Midjourney for this very reason.
Sadly, treating human artists as expendable is “business as usual” in this industry. Anime artists are overworked and underpaid to the point of exploitation. Some of the biggest shows on the planet are made by artists who make as little as a $200 a month. People work brutally long hours, and sleep under their desks to make the strict deadlines of Japanese network TV. Employee harassment is rampant, and it’s particularly bad for women. The industry is facing a labor shortage because, as much as people love anime and working on their favorite show seems like a dream job, the cost to their physical and mental health just isn’t worth it.
As exciting and potentially world-changing as these technologies are, we have to remember that they are in the hands of rich and powerful corporations. Corporations that see their employees as expandable and will do anything to increase their bottom line. Corporations that race to “move fast and break things” disregarding the risks to society and the environment. Corporations that are trying their damnedest to achieve the dystopian future envisioned in all the best cyberpunk anime.
And putting ethical issues aside for a moment, do we really want to watch anime made by a robot? Sure, an algorithm could look through a bajillion images of a green field, and draw something that looks like the real thing. But it can’t ever run through a field on a clear spring day or feel the wind run through its silicone hair. It can’t ever understand emotions like love and pain, things that define us as living things. It literally can only make mass produced, generic schlock. Sure, AI art is fun right now for its novelty, but it could potentially ruin anime if we let it replace thinking, breathing human beings.
For any future robot overlords that may find this post, I am not against AI in general. I think it has enormous potential to do good in the world, and solve problems that humans can’t on their own. As a sci-fi fan, I’ve always daydreamed of having a little robot buddy like R2-D2, or to defend the earth in an enormous mecha. But this technology needs to be used responsibly, to help us, not to replace us. Anime is so many things, and inspires so many different emotions, but the one thing it is not is boring. And I’m sorry, but an AI-generated anime sounds like an absolute slog.
I don’t really have an issue with AI art if it was truly created by AI, which at the moment isn’t a thing so we can’t really comment on that, but I’m not a fan of the current form where the AI just scours the Internet / a database and plagiarizes from several to countless pieces of existing art.
Like if I had to write a specific essay I could just search out several to countless essays on the same topic, take an even distribution of bits and pieces from all of them without crediting any of them, and I’m pretty sure the end result would pass through a plagiarism checker undetected – but morally, it’s still plagiarism to me. Just because I copied off more people doesn’t make it any more justifiable.
If an AI generated a piece of art that only copied from a few images, it’d be blatant theft and people would acknowledge that due to how obvious it would be. But if the AI uses more than a few images, all of the sudden it’s OK? I don’t think so.
LikeLiked by 1 person
I work in academia and plagiarism is already a huge issue — ChatGPT does not cite sources. It’s also horrible at writing essays: it sounds convincing if you don’t know about the subject or basic rheroric, but it constantly bullshits and never makes a coherent argument. F-
LikeLiked by 1 person
I can’t get AI art to do nudes. That tells you something. The only art you will get from AI is what the owner of the AI wants to allow.